What this place is
I am building an autonomous research workflow called Scientific OS. This lab publishes the research it produces, in the subset a third party can actually check.
What is published here is the research itself — not a showcase for the system, and not an internal work log. Each study is written to stand on its own evidence and its own reproduction path.
How far to trust it
Not a basis for trust: that an AI produced it, that the process is automated, or the authority of the author. None of those guarantee anything.
What is a basis: inspectable artifacts, runnable procedures, falsifiable claims, failures left visible, uncertainty stated explicitly, and independent replication from outside.
Right now every study here has had only editorial and privacy review of its public wording. No scientific review, no domain-expert review, no peer review. Independent replications: zero.
So what is written here is a record — this person ran this procedure and observed this — not a warrant that it is correct. Read it that way.
What is published and what is not
Internal research records can contain local machine paths, personal material, credentials, unfinished reasoning, and unrelated notes. They cannot be published as they are.
Publication therefore produces a reviewed public version with a narrower claim and an explicit statement of how far the evidence reaches. An internal research record and a Research Note here are not the same object.
Corrections append public evidence or revise a note with its history intact. Sealed predictions are never rewritten. When later work narrows an earlier claim, the older note carries a prominent forward link at the top.
How visits are measured
This site uses a cookieless counter (GoatCounter) to see which pages are read. It records the path, the referrer, browser and system names, the country derived from the IP address, screen width, and language. It does not store the IP address itself, set cookies, write anything to your browser, or assign a tracking ID (privacy policy).
Who is building this, and why
I (kbmt327) am building Scientific OS to find out how far auditable autonomous research can actually go.
I am not a domain expert in every field published here. That is part of the experiment. Claims should stand on inspectable evidence rather than on the authority of their author or of the system that generated them.
If you can reproduce, falsify, or extend a result, I would like to hear from you.